summaryrefslogtreecommitdiff
path: root/lib/xz/xz_dec_lzma2.c
blob: 9f336bc07ed610f6944e6617439dd7fd0e51c9e0 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
/*
 * LZMA2 decoder
 *
 * Authors: Lasse Collin <lasse.collin@tukaani.org>
 *          Igor Pavlov <https://7-zip.org/>
 *
 * This file has been put into the public domain.
 * You can do whatever you want with this file.
 */

#include "xz_private.h"
#include "xz_lzma2.h"

/*
 * Range decoder initialization eats the first five bytes of each LZMA chunk.
 */
#define RC_INIT_BYTES 5

/*
 * Minimum number of usable input buffer to safely decode one LZMA symbol.
 * The worst case is that we decode 22 bits using probabilities and 26
 * direct bits. This may decode at maximum of 20 bytes of input. However,
 * lzma_main() does an extra normalization before returning, thus we
 * need to put 21 here.
 */
#define LZMA_IN_REQUIRED 21

/*
 * Dictionary (history buffer)
 *
 * These are always true:
 *    start <= pos <= full <= end
 *    pos <= limit <= end
 *
 * In multi-call mode, also these are true:
 *    end == size
 *    size <= size_max
 *    allocated <= size
 *
 * Most of these variables are size_t to support single-call mode,
 * in which the dictionary variables address the actual output
 * buffer directly.
 */
struct dictionary {
	/* Beginning of the history buffer */
	uint8_t *buf;

	/* Old position in buf (before decoding more data) */
	size_t start;

	/* Position in buf */
	size_t pos;

	/*
	 * How full dictionary is. This is used to detect corrupt input that
	 * would read beyond the beginning of the uncompressed stream.
	 */
	size_t full;

	/* Write limit; we don't write to buf[limit] or later bytes. */
	size_t limit;

	/*
	 * End of the dictionary buffer. In multi-call mode, this is
	 * the same as the dictionary size. In single-call mode, this
	 * indicates the size of the output buffer.
	 */
	size_t end;

	/*
	 * Size of the dictionary as specified in Block Header. This is used
	 * together with "full" to detect corrupt input that would make us
	 * read beyond the beginning of the uncompressed stream.
	 */
	uint32_t size;

	/*
	 * Maximum allowed dictionary size in multi-call mode.
	 * This is ignored in single-call mode.
	 */
	uint32_t size_max;

	/*
	 * Amount of memory currently allocated for the dictionary.
	 * This is used only with XZ_DYNALLOC. (With XZ_PREALLOC,
	 * size_max is always the same as the allocated size.)
	 */
	uint32_t allocated;

	/* Operation mode */
	enum xz_mode mode;
};

/* Range decoder */
struct rc_dec {
	uint32_t range;
	uint32_t code;

	/*
	 * Number of initializing bytes remaining to be read
	 * by rc_read_init().
	 */
	uint32_t init_bytes_left;

	/*
	 * Buffer from which we read our input. It can be either
	 * temp.buf or the caller-provided input buffer.
	 */
	const uint8_t *in;
	size_t in_pos;
	size_t in_limit;
};

/* Probabilities for a length decoder. */
struct lzma_len_dec {
	/* Probability of match length being at least 10 */
	uint16_t choice;

	/* Probability of match length being at least 18 */
	uint16_t choice2;

	/* Probabilities for match lengths 2-9 */
	uint16_t low[POS_STATES_MAX][LEN_LOW_SYMBOLS];

	/* Probabilities for match lengths 10-17 */
	uint16_t mid[POS_STATES_MAX][LEN_MID_SYMBOLS];

	/* Probabilities for match lengths 18-273 */
	uint16_t high[LEN_HIGH_SYMBOLS];
};

struct lzma_dec {
	/* Distances of latest four matches */
	uint32_t rep0;
	uint32_t rep1;
	uint32_t rep2;
	uint32_t rep3;

	/* Types of the most recently seen LZMA symbols */
	enum lzma_state state;

	/*
	 * Length of a match. This is updated so that dict_repeat can
	 * be called again to finish repeating the whole match.
	 */
	uint32_t len;

	/*
	 * LZMA properties or related bit masks (number of literal
	 * context bits, a mask dervied from the number of literal
	 * position bits, and a mask dervied from the number
	 * position bits)
	 */
	uint32_t lc;
	uint32_t literal_pos_mask; /* (1 << lp) - 1 */
	uint32_t pos_mask;         /* (1 << pb) - 1 */

	/* If 1, it's a match. Otherwise it's a single 8-bit literal. */
	uint16_t is_match[STATES][POS_STATES_MAX];

	/* If 1, it's a repeated match. The distance is one of rep0 .. rep3. */
	uint16_t is_rep[STATES];

	/*
	 * If 0, distance of a repeated match is rep0.
	 * Otherwise check is_rep1.
	 */
	uint16_t is_rep0[STATES];

	/*
	 * If 0, distance of a repeated match is rep1.
	 * Otherwise check is_rep2.
	 */
	uint16_t is_rep1[STATES];

	/* If 0, distance of a repeated match is rep2. Otherwise it is rep3. */
	uint16_t is_rep2[STATES];

	/*
	 * If 1, the repeated match has length of one byte. Otherwise
	 * the length is decoded from rep_len_decoder.
	 */
	uint16_t is_rep0_long[STATES][POS_STATES_MAX];

	/*
	 * Probability tree for the highest two bits of the match
	 * distance. There is a separate probability tree for match
	 * lengths of 2 (i.e. MATCH_LEN_MIN), 3, 4, and [5, 273].
	 */
	uint16_t dist_slot[DIST_STATES][DIST_SLOTS];

	/*
	 * Probility trees for additional bits for match distance
	 * when the distance is in the range [4, 127].
	 */
	uint16_t dist_special[FULL_DISTANCES - DIST_MODEL_END];

	/*
	 * Probability tree for the lowest four bits of a match
	 * distance that is equal to or greater than 128.
	 */
	uint16_t dist_align[ALIGN_SIZE];

	/* Length of a normal match */
	struct lzma_len_dec match_len_dec;

	/* Length of a repeated match */
	struct lzma_len_dec rep_len_dec;

	/* Probabilities of literals */
	uint16_t literal[LITERAL_CODERS_MAX][LITERAL_CODER_SIZE];
};

struct lzma2_dec {
	/* Position in xz_dec_lzma2_run(). */
	enum lzma2_seq {
		SEQ_CONTROL,
		SEQ_UNCOMPRESSED_1,
		SEQ_UNCOMPRESSED_2,
		SEQ_COMPRESSED_0,
		SEQ_COMPRESSED_1,
		SEQ_PROPERTIES,
		SEQ_LZMA_PREPARE,
		SEQ_LZMA_RUN,
		SEQ_COPY
	} sequence;

	/* Next position after decoding the compressed size of the chunk. */
	enum lzma2_seq next_sequence;

	/* Uncompressed size of LZMA chunk (2 MiB at maximum) */
	uint32_t uncompressed;

	/*
	 * Compressed size of LZMA chunk or compressed/uncompressed
	 * size of uncompressed chunk (64 KiB at maximum)
	 */
	uint32_t compressed;

	/*
	 * True if dictionary reset is needed. This is false before
	 * the first chunk (LZMA or uncompressed).
	 */
	bool need_dict_reset;

	/*
	 * True if new LZMA properties are needed. This is false
	 * before the first LZMA chunk.
	 */
	bool need_props;
};

struct xz_dec_lzma2 {
	/*
	 * The order below is important on x86 to reduce code size and
	 * it shouldn't hurt on other platforms. Everything up to and
	 * including lzma.pos_mask are in the first 128 bytes on x86-32,
	 * which allows using smaller instructions to access those
	 * variables. On x86-64, fewer variables fit into the first 128
	 * bytes, but this is still the best order without sacrificing
	 * the readability by splitting the structures.
	 */
	struct rc_dec rc;
	struct dictionary dict;
	struct lzma2_dec lzma2;
	struct lzma_dec lzma;

	/*
	 * Temporary buffer which holds small number of input bytes between
	 * decoder calls. See lzma2_lzma() for details.
	 */
	struct {
		uint32_t size;
		uint8_t buf[3 * LZMA_IN_REQUIRED];
	} temp;
};

/**************
 * Dictionary *
 **************/

/*
 * Reset the dictionary state. When in single-call mode, set up the beginning
 * of the dictionary to point to the actual output buffer.
 */
static void dict_reset(struct dictionary *dict, struct xz_buf *b)
{
	if (DEC_IS_SINGLE(dict->mode)) {
		dict->buf = b->out + b->out_pos;
		dict->end = b->out_size - b->out_pos;
	}

	dict->start = 0;
	dict->pos = 0;
	dict->limit = 0;
	dict->full = 0;
}

/* Set dictionary write limit */
static void dict_limit(struct dictionary *dict, size_t out_max)
{
	if (dict->end - dict->pos <= out_max)
		dict->limit = dict->end;
	else
		dict->limit = dict->pos + out_max;
}

/* Return true if at least one byte can be written into the dictionary. */
static inline bool dict_has_space(const struct dictionary *dict)
{
	return dict->pos < dict->limit;
}

/*
 * Get a byte from the dictionary at the given distance. The distance is
 * assumed to valid, or as a special case, zero when the dictionary is
 * still empty. This special case is needed for single-call decoding to
 * avoid writing a '\0' to the end of the destination buffer.
 */
static inline uint32_t dict_get(const struct dictionary *dict, uint32_t dist)
{
	size_t offset = dict->pos - dist - 1;

	if (dist >= dict->pos)
		offset += dict->end;

	return dict->full > 0 ? dict->buf[offset] : 0;
}

/*
 * Put one byte into the dictionary. It is assumed that there is space for it.
 */
static inline void dict_put(struct dictionary *dict, uint8_t byte)
{
	dict->buf[dict->pos++] = byte;

	if (dict->full < dict->pos)
		dict->full = dict->pos;
}

/*
 * Repeat given number of bytes from the given distance. If the distance is
 * invalid, false is returned. On success, true is returned and *len is
 * updated to indicate how many bytes were left to be repeated.
 */
static bool dict_repeat(struct dictionary *dict, uint32_t *len, uint32_t dist)
{
	size_t back;
	uint32_t left;

	if (dist >= dict->full || dist >= dict->size)
		return false;

	left = min_t(size_t, dict->limit - dict->pos, *len);
	*len -= left;

	back = dict->pos - dist - 1;
	if (dist >= dict->pos)
		back += dict->end;

	do {
		dict->buf[dict->pos++] = dict->buf[back++];
		if (back == dict->end)
			back = 0;
	} while (--left > 0);

	if (dict->full < dict->pos)
		dict->full = dict->pos;

	return true;
}

/* Copy uncompressed data as is from input to dictionary and output buffers. */
static void dict_uncompressed(struct dictionary *dict, struct xz_buf *b,
			      uint32_t *left)
{
	size_t copy_size;

	while (*left > 0 && b->in_pos < b->in_size
			&& b->out_pos < b->out_size) {
		copy_size = min(b->in_size - b->in_pos,
				b->out_size - b->out_pos);
		if (copy_size > dict->end - dict->pos)
			copy_size = dict->end - dict->pos;
		if (copy_size > *left)
			copy_size = *left;

		*left -= copy_size;

		memcpy(dict->buf + dict->pos, b->in + b->in_pos, copy_size);
		dict->pos += copy_size;

		if (dict->full < dict->pos)
			dict->full = dict->pos;

		if (DEC_IS_MULTI(dict->mode)) {
			if (dict->pos == dict->end)
				dict->pos = 0;

			memcpy(b->out + b->out_pos, b->in + b->in_pos,
					copy_size);
		}

		dict->start = dict->pos;

		b->out_pos += copy_size;
		b->in_pos += copy_size;
	}
}

/*
 * Flush pending data from dictionary to b->out. It is assumed that there is
 * enough space in b->out. This is guaranteed because caller uses dict_limit()
 * before decoding data into the dictionary.
 */
static uint32_t dict_flush(struct dictionary *dict, struct xz_buf *b)
{
	size_t copy_size = dict->pos - dict->start;

	if (DEC_IS_MULTI(dict->mode)) {
		if (dict->pos == dict->end)
			dict->pos = 0;

		memcpy(b->out + b->out_pos, dict->buf + dict->start,
				copy_size);
	}

	dict->start = dict->pos;
	b->out_pos += copy_size;
	return copy_size;
}

/*****************
 * Range decoder *
 *****************/

/* Reset the range decoder. */
static void rc_reset(struct rc_dec *rc)
{
	rc->range = (uint32_t)-1;
	rc->code = 0;
	rc->init_bytes_left = RC_INIT_BYTES;
}

/*
 * Read the first five initial bytes into rc->code if they haven't been
 * read already. (Yes, the first byte gets completely ignored.)
 */
static bool rc_read_init(struct rc_dec *rc, struct xz_buf *b)
{
	while (rc->init_bytes_left > 0) {
		if (b->in_pos == b->in_size)
			return false;

		rc->code = (rc->code << 8) + b->in[b->in_pos++];
		--rc->init_bytes_left;
	}

	return true;
}

/* Return true if there may not be enough input for the next decoding loop. */
static inline bool rc_limit_exceeded(const struct rc_dec *rc)
{
	return rc->in_pos > rc->in_limit;
}

/*
 * Return true if it is possible (from point of view of range decoder) that
 * we have reached the end of the LZMA chunk.
 */
static inline bool rc_is_finished(const struct rc_dec *rc)
{
	return rc->code == 0;
}

/* Read the next input byte if needed. */
static __always_inline void rc_normalize(struct rc_dec *rc)
{
	if (rc->range < RC_TOP_VALUE) {
		rc->range <<= RC_SHIFT_BITS;
		rc->code = (rc->code << RC_SHIFT_BITS) + rc->in[rc->in_pos++];
	}
}

/*
 * Decode one bit. In some versions, this function has been splitted in three
 * functions so that the compiler is supposed to be able to more easily avoid
 * an extra branch. In this particular version of the LZMA decoder, this
 * doesn't seem to be a good idea (tested with GCC 3.3.6, 3.4.6, and 4.3.3
 * on x86). Using a non-splitted version results in nicer looking code too.
 *
 * NOTE: This must return an int. Do not make it return a bool or the speed
 * of the code generated by GCC 3.x decreases 10-15 %. (GCC 4.3 doesn't care,
 * and it generates 10-20 % faster code than GCC 3.x from this file anyway.)
 */
static __always_inline int rc_bit(struct rc_dec *rc, uint16_t *prob)
{
	uint32_t bound;
	int bit;

	rc_normalize(rc);
	bound = (rc->range >> RC_BIT_MODEL_TOTAL_BITS) * *prob;
	if (rc->code < bound) {
		rc->range = bound;
		*prob += (RC_BIT_MODEL_TOTAL - *prob) >> RC_MOVE_BITS;
		bit = 0;
	} else {
		rc->range -= bound;
		rc->code -= bound;
		*prob -= *prob >> RC_MOVE_BITS;
		bit = 1;
	}

	return bit;
}

/* Decode a bittree starting from the most significant bit. */
static __always_inline uint32_t rc_bittree(struct rc_dec *rc,
					   uint16_t *probs, uint32_t limit)
{
	uint32_t symbol = 1;

	do {
		if (rc_bit(rc, &probs[symbol]))
			symbol = (symbol << 1) + 1;
		else
			symbol <<= 1;
	} while (symbol < limit);

	return symbol;
}

/* Decode a bittree starting from the least significant bit. */
static __always_inline void rc_bittree_reverse(struct rc_dec *rc,
					       uint16_t *probs,
					       uint32_t *dest, uint32_t limit)
{
	uint32_t symbol = 1;
	uint32_t i = 0;

	do {
		if (rc_bit(rc, &probs[symbol])) {
			symbol = (symbol << 1) + 1;
			*dest += 1 << i;
		} else {
			symbol <<= 1;
		}
	} while (++i < limit);
}

/* Decode direct bits (fixed fifty-fifty probability) */
static inline void rc_direct(struct rc_dec *rc, uint32_t *dest, uint32_t limit)
{
	uint32_t mask;

	do {
		rc_normalize(rc);
		rc->range >>= 1;
		rc->code -= rc->range;
		mask = (uint32_t)0 - (rc->code >> 31);
		rc->code += rc->range & mask;
		*dest = (*dest << 1) + (mask + 1);
	} while (--limit > 0);
}

/********
 * LZMA *
 ********/

/* Get pointer to literal coder probability array. */
static uint16_t *lzma_literal_probs(struct xz_dec_lzma2 *s)
{
	uint32_t prev_byte = dict_get(&s->dict, 0);
	uint32_t low = prev_byte >> (8 - s->lzma.lc);
	uint32_t high = (s->dict.pos & s->lzma.literal_pos_mask) << s->lzma.lc;
	return s->lzma.literal[low + high];
}

/* Decode a literal (one 8-bit byte) */
static void lzma_literal(struct xz_dec_lzma2 *s)
{
	uint16_t *probs;
	uint32_t symbol;
	uint32_t match_byte;
	uint32_t match_bit;
	uint32_t offset;
	uint32_t i;

	probs = lzma_literal_probs(s);

	if (lzma_state_is_literal(s->lzma.state)) {
		symbol = rc_bittree(&s->rc, probs, 0x100);
	} else {
		symbol = 1;
		match_byte = dict_get(&s->dict, s->lzma.rep0) << 1;
		offset = 0x100;

		do {
			match_bit = match_byte & offset;
			match_byte <<= 1;
			i = offset + match_bit + symbol;

			if (rc_bit(&s->rc, &probs[i])) {
				symbol = (symbol << 1) + 1;
				offset &= match_bit;
			} else {
				symbol <<= 1;
				offset &= ~match_bit;
			}
		} while (symbol < 0x100);
	}

	dict_put(&s->dict, (uint8_t)symbol);
	lzma_state_literal(&s->lzma.state);
}

/* Decode the length of the match into s->lzma.len. */
static void lzma_len(struct xz_dec_lzma2 *s, struct lzma_len_dec *l,
		     uint32_t pos_state)
{
	uint16_t *probs;
	uint32_t limit;

	if (!rc_bit(&s->rc, &l->choice)) {
		probs = l->low[pos_state];
		limit = LEN_LOW_SYMBOLS;
		s->lzma.len = MATCH_LEN_MIN;
	} else {
		if (!rc_bit(&s->rc, &l->choice2)) {
			probs = l->mid[pos_state];
			limit = LEN_MID_SYMBOLS;
			s->lzma.len = MATCH_LEN_MIN + LEN_LOW_SYMBOLS;
		} else {
			probs = l->high;
			limit = LEN_HIGH_SYMBOLS;
			s->lzma.len = MATCH_LEN_MIN + LEN_LOW_SYMBOLS
					+ LEN_MID_SYMBOLS;
		}
	}

	s->lzma.len += rc_bittree(&s->rc, probs, limit) - limit;
}

/* Decode a match. The distance will be stored in s->lzma.rep0. */
static void lzma_match(struct xz_dec_lzma2 *s, uint32_t pos_state)
{
	uint16_t *probs;
	uint32_t dist_slot;
	uint32_t limit;

	lzma_state_match(&s->lzma.state);

	s->lzma.rep3 = s->lzma.rep2;
	s->lzma.rep2 = s->lzma.rep1;
	s->lzma.rep1 = s->lzma.rep0;

	lzma_len(s, &s->lzma.match_len_dec, pos_state);

	probs = s->lzma.dist_slot[lzma_get_dist_state(s->lzma.len)];
	dist_slot = rc_bittree(&s->rc, probs, DIST_SLOTS) - DIST_SLOTS;

	if (dist_slot < DIST_MODEL_START) {
		s->lzma.rep0 = dist_slot;
	} else {
		limit = (dist_slot >> 1) - 1;
		s->lzma.rep0 = 2 + (dist_slot & 1);

		if (dist_slot < DIST_MODEL_END) {
			s->lzma.rep0 <<= limit;
			probs = s->lzma.dist_special + s->lzma.rep0
					- dist_slot - 1;
			rc_bittree_reverse(&s->rc, probs,
					&s->lzma.rep0, limit);
		} else {
			rc_direct(&s->rc, &s->lzma.rep0, limit - ALIGN_BITS);
			s->lzma.rep0 <<= ALIGN_BITS;
			rc_bittree_reverse(&s->rc, s->lzma.dist_align,
					&s->lzma.rep0, ALIGN_BITS);
		}
	}
}

/*
 * Decode a repeated match. The distance is one of the four most recently
 * seen matches. The distance will be stored in s->lzma.rep0.
 */
static void lzma_rep_match(struct xz_dec_lzma2 *s, uint32_t pos_state)
{
	uint32_t tmp;

	if (!rc_bit(&s->rc, &s->lzma.is_rep0[s->lzma.state])) {
		if (!rc_bit(&s->rc, &s->lzma.is_rep0_long[
				s->lzma.state][pos_state])) {
			lzma_state_short_rep(&s->lzma.state);
			s->lzma.len = 1;
			return;
		}
	} else {
		if (!rc_bit(&s->rc, &s->lzma.is_rep1[s->lzma.state])) {
			tmp = s->lzma.rep1;
		} else {
			if (!rc_bit(&s->rc, &s->lzma.is_rep2[s->lzma.state])) {
				tmp = s->lzma.rep2;
			} else {
				tmp = s->lzma.rep3;
				s->lzma.rep3 = s->lzma.rep2;
			}

			s->lzma.rep2 = s->lzma.rep1;
		}

		s->lzma.rep1 = s->lzma.rep0;
		s->lzma.rep0 = tmp;
	}

	lzma_state_long_rep(&s->lzma.state);
	lzma_len(s, &s->lzma.rep_len_dec, pos_state);
}

/* LZMA decoder core */
static bool lzma_main(struct xz_dec_lzma2 *s)
{
	uint32_t pos_state;

	/*
	 * If the dictionary was reached during the previous call, try to
	 * finish the possibly pending repeat in the dictionary.
	 */
	if (dict_has_space(&s->dict) && s->lzma.len > 0)
		dict_repeat(&s->dict, &s->lzma.len, s->lzma.rep0);

	/*
	 * Decode more LZMA symbols. One iteration may consume up to
	 * LZMA_IN_REQUIRED - 1 bytes.
	 */
	while (dict_has_space(&s->dict) && !rc_limit_exceeded(&s->rc)) {
		pos_state = s->dict.pos & s->lzma.pos_mask;

		if (!rc_bit(&s->rc, &s->lzma.is_match[
				s->lzma.state][pos_state])) {
			lzma_literal(s);
		} else {
			if (rc_bit(&s->rc, &s->lzma.is_rep[s->lzma.state]))
				lzma_rep_match(s, pos_state);
			else
				lzma_match(s, pos_state);

			if (!dict_repeat(&s->dict, &s->lzma.len, s->lzma.rep0))
				return false;
		}
	}

	/*
	 * Having the range decoder always normalized when we are outside
	 * this function makes it easier to correctly handle end of the chunk.
	 */
	rc_normalize(&s->rc);

	return true;
}

/*
 * Reset the LZMA decoder and range decoder state. Dictionary is nore reset
 * here, because LZMA state may be reset without resetting the dictionary.
 */
static void lzma_reset(struct xz_dec_lzma2 *s)
{
	uint16_t *probs;
	size_t i;

	s->lzma.state = STATE_LIT_LIT;
	s->lzma.rep0 = 0;
	s->lzma.rep1 = 0;
	s->lzma.rep2 = 0;
	s->lzma.rep3 = 0;

	/*
	 * All probabilities are initialized to the same value. This hack
	 * makes the code smaller by avoiding a separate loop for each
	 * probability array.
	 *
	 * This could be optimized so that only that part of literal
	 * probabilities that are actually required. In the common case
	 * we would write 12 KiB less.
	 */
	probs = s->lzma.is_match[0];
	for (i = 0; i < PROBS_TOTAL; ++i)
		probs[i] = RC_BIT_MODEL_TOTAL / 2;

	rc_reset(&s->rc);
}

/*
 * Decode and validate LZMA properties (lc/lp/pb) and calculate the bit masks
 * from the decoded lp and pb values. On success, the LZMA decoder state is
 * reset and true is returned.
 */
static bool lzma_props(struct xz_dec_lzma2 *s, uint8_t props)
{
	if (props > (4 * 5 + 4) * 9 + 8)
		return false;

	s->lzma.pos_mask = 0;
	while (props >= 9 * 5) {
		props -= 9 * 5;
		++s->lzma.pos_mask;
	}

	s->lzma.pos_mask = (1 << s->lzma.pos_mask) - 1;

	s->lzma.literal_pos_mask = 0;
	while (props >= 9) {
		props -= 9;
		++s->lzma.literal_pos_mask;
	}

	s->lzma.lc = props;

	if (s->lzma.lc + s->lzma.literal_pos_mask > 4)
		return false;

	s->lzma.literal_pos_mask = (1 << s->lzma.literal_pos_mask) - 1;

	lzma_reset(s);

	return true;
}

/*********
 * LZMA2 *
 *********/

/*
 * The LZMA decoder assumes that if the input limit (s->rc.in_limit) hasn't
 * been exceeded, it is safe to read up to LZMA_IN_REQUIRED bytes. This
 * wrapper function takes care of making the LZMA decoder's assumption safe.
 *
 * As long as there is plenty of input left to be decoded in the current LZMA
 * chunk, we decode directly from the caller-supplied input buffer until
 * there's LZMA_IN_REQUIRED bytes left. Those remaining bytes are copied into
 * s->temp.buf, which (hopefully) gets filled on the next call to this
 * function. We decode a few bytes from the temporary buffer so that we can
 * continue decoding from the caller-supplied input buffer again.
 */
static bool lzma2_lzma(struct xz_dec_lzma2 *s, struct xz_buf *b)
{
	size_t in_avail;
	uint32_t tmp;

	in_avail = b->in_size - b->in_pos;
	if (s->temp.size > 0 || s->lzma2.compressed == 0) {
		tmp = 2 * LZMA_IN_REQUIRED - s->temp.size;
		if (tmp > s->lzma2.compressed - s->temp.size)
			tmp = s->lzma2.compressed - s->temp.size;
		if (tmp > in_avail)
			tmp = in_avail;

		memcpy(s->temp.buf + s->temp.size, b->in + b->in_pos, tmp);

		if (s->temp.size + tmp == s->lzma2.compressed) {
			memzero(s->temp.buf + s->temp.size + tmp,
					sizeof(s->temp.buf)
						- s->temp.size - tmp);
			s->rc.in_limit = s->temp.size + tmp;
		} else if (s->temp.size + tmp < LZMA_IN_REQUIRED) {
			s->temp.size += tmp;
			b->in_pos += tmp;
			return true;
		} else {
			s->rc.in_limit = s->temp.size + tmp - LZMA_IN_REQUIRED;
		}

		s->rc.in = s->temp.buf;
		s->rc.in_pos = 0;

		if (!lzma_main(s) || s->rc.in_pos > s->temp.size + tmp)
			return false;

		s->lzma2.compressed -= s->rc.in_pos;

		if (s->rc.in_pos < s->temp.size) {
			s->temp.size -= s->rc.in_pos;
			memmove(s->temp.buf, s->temp.buf + s->rc.in_pos,
					s->temp.size);
			return true;
		}

		b->in_pos += s->rc.in_pos - s->temp.size;
		s->temp.size = 0;
	}

	in_avail = b->in_size - b->in_pos;
	if (in_avail >= LZMA_IN_REQUIRED) {
		s->rc.in = b->in;
		s->rc.in_pos = b->in_pos;

		if (in_avail >= s->lzma2.compressed + LZMA_IN_REQUIRED)
			s->rc.in_limit = b->in_pos + s->lzma2.compressed;
		else
			s->rc.in_limit = b->in_size - LZMA_IN_REQUIRED;

		if (!lzma_main(s))
			return false;

		in_avail = s->rc.in_pos - b->in_pos;
		if (in_avail > s->lzma2.compressed)
			return false;

		s->lzma2.compressed -= in_avail;
		b->in_pos = s->rc.in_pos;
	}

	in_avail = b->in_size - b->in_pos;
	if (in_avail < LZMA_IN_REQUIRED) {
		if (in_avail > s->lzma2.compressed)
			in_avail = s->lzma2.compressed;

		memcpy(s->temp.buf, b->in + b->in_pos, in_avail);
		s->temp.size = in_avail;
		b->in_pos += in_avail;
	}

	return true;
}

/*
 * Take care of the LZMA2 control layer, and forward the job of actual LZMA
 * decoding or copying of uncompressed chunks to other functions.
 */
XZ_EXTERN enum xz_ret xz_dec_lzma2_run(struct xz_dec_lzma2 *s,
				       struct xz_buf *b)
{
	uint32_t tmp;

	while (b->in_pos < b->in_size || s->lzma2.sequence == SEQ_LZMA_RUN) {
		switch (s->lzma2.sequence) {
		case SEQ_CONTROL:
			/*
			 * LZMA2 control byte
			 *
			 * Exact values:
			 *   0x00   End marker
			 *   0x01   Dictionary reset followed by
			 *          an uncompressed chunk
			 *   0x02   Uncompressed chunk (no dictionary reset)
			 *
			 * Highest three bits (s->control & 0xE0):
			 *   0xE0   Dictionary reset, new properties and state
			 *          reset, followed by LZMA compressed chunk
			 *   0xC0   New properties and state reset, followed
			 *          by LZMA compressed chunk (no dictionary
			 *          reset)
			 *   0xA0   State reset using old properties,
			 *          followed by LZMA compressed chunk (no
			 *          dictionary reset)
			 *   0x80   LZMA chunk (no dictionary or state reset)
			 *
			 * For LZMA compressed chunks, the lowest five bits
			 * (s->control & 1F) are the highest bits of the
			 * uncompressed size (bits 16-20).
			 *
			 * A new LZMA2 stream must begin with a dictionary
			 * reset. The first LZMA chunk must set new
			 * properties and reset the LZMA state.
			 *
			 * Values that don't match anything described above
			 * are invalid and we return XZ_DATA_ERROR.
			 */
			tmp = b->in[b->in_pos++];

			if (tmp == 0x00)
				return XZ_STREAM_END;

			if (tmp >= 0xE0 || tmp == 0x01) {
				s->lzma2.need_props = true;
				s->lzma2.need_dict_reset = false;
				dict_reset(&s->dict, b);
			} else if (s->lzma2.need_dict_reset) {
				return XZ_DATA_ERROR;
			}

			if (tmp >= 0x80) {
				s->lzma2.uncompressed = (tmp & 0x1F) << 16;
				s->lzma2.sequence = SEQ_UNCOMPRESSED_1;

				if (tmp >= 0xC0) {
					/*
					 * When there are new properties,
					 * state reset is done at
					 * SEQ_PROPERTIES.
					 */
					s->lzma2.need_props = false;
					s->lzma2.next_sequence
							= SEQ_PROPERTIES;

				} else if (s->lzma2.need_props) {
					return XZ_DATA_ERROR;

				} else {
					s->lzma2.next_sequence
							= SEQ_LZMA_PREPARE;
					if (tmp >= 0xA0)
						lzma_reset(s);
				}
			} else {
				if (tmp > 0x02)
					return XZ_DATA_ERROR;

				s->lzma2.sequence = SEQ_COMPRESSED_0;
				s->lzma2.next_sequence = SEQ_COPY;
			}

			break;

		case SEQ_UNCOMPRESSED_1:
			s->lzma2.uncompressed
					+= (uint32_t)b->in[b->in_pos++] << 8;
			s->lzma2.sequence = SEQ_UNCOMPRESSED_2;
			break;

		case SEQ_UNCOMPRESSED_2:
			s->lzma2.uncompressed
					+= (uint32_t)b->in[b->in_pos++] + 1;
			s->lzma2.sequence = SEQ_COMPRESSED_0;
			break;

		case SEQ_COMPRESSED_0:
			s->lzma2.compressed
					= (uint32_t)b->in[b->in_pos++] << 8;
			s->lzma2.sequence = SEQ_COMPRESSED_1;
			break;

		case SEQ_COMPRESSED_1:
			s->lzma2.compressed
					+= (uint32_t)b->in[b->in_pos++] + 1;
			s->lzma2.sequence = s->lzma2.next_sequence;
			break;

		case SEQ_PROPERTIES:
			if (!lzma_props(s, b->in[b->in_pos++]))
				return XZ_DATA_ERROR;

			s->lzma2.sequence = SEQ_LZMA_PREPARE;

		/* Fall through */

		case SEQ_LZMA_PREPARE:
			if (s->lzma2.compressed < RC_INIT_BYTES)
				return XZ_DATA_ERROR;

			if (!rc_read_init(&s->rc, b))
				return XZ_OK;

			s->lzma2.compressed -= RC_INIT_BYTES;
			s->lzma2.sequence = SEQ_LZMA_RUN;

		/* Fall through */

		case SEQ_LZMA_RUN:
			/*
			 * Set dictionary limit to indicate how much we want
			 * to be encoded at maximum. Decode new data into the
			 * dictionary. Flush the new data from dictionary to
			 * b->out. Check if we finished decoding this chunk.
			 * In case the dictionary got full but we didn't fill
			 * the output buffer yet, we may run this loop
			 * multiple times without changing s->lzma2.sequence.
			 */
			dict_limit(&s->dict, min_t(size_t,
					b->out_size - b->out_pos,
					s->lzma2.uncompressed));
			if (!lzma2_lzma(s, b))
				return XZ_DATA_ERROR;

			s->lzma2.uncompressed -= dict_flush(&s->dict, b);

			if (s->lzma2.uncompressed == 0) {
				if (s->lzma2.compressed > 0 || s->lzma.len > 0
						|| !rc_is_finished(&s->rc))
					return XZ_DATA_ERROR;

				rc_reset(&s->rc);
				s->lzma2.sequence = SEQ_CONTROL;

			} else if (b->out_pos == b->out_size
					|| (b->in_pos == b->in_size
						&& s->temp.size
						< s->lzma2.compressed)) {
				return XZ_OK;
			}

			break;

		case SEQ_COPY:
			dict_uncompressed(&s->dict, b, &s->lzma2.compressed);
			if (s->lzma2.compressed > 0)
				return XZ_OK;

			s->lzma2.sequence = SEQ_CONTROL;
			break;
		}
	}

	return XZ_OK;
}

XZ_EXTERN struct xz_dec_lzma2 *xz_dec_lzma2_create(enum xz_mode mode,
						   uint32_t dict_max)
{
	struct xz_dec_lzma2 *s = kmalloc(sizeof(*s), GFP_KERNEL);
	if (s == NULL)
		return NULL;

	s->dict.mode = mode;
	s->dict.size_max = dict_max;

	if (DEC_IS_PREALLOC(mode)) {
		s->dict.buf = vmalloc(dict_max);
		if (s->dict.buf == NULL) {
			kfree(s);
			return NULL;
		}
	} else if (DEC_IS_DYNALLOC(mode)) {
		s->dict.buf = NULL;
		s->dict.allocated = 0;
	}

	return s;
}

XZ_EXTERN enum xz_ret xz_dec_lzma2_reset(struct xz_dec_lzma2 *s, uint8_t props)
{
	/* This limits dictionary size to 3 GiB to keep parsing simpler. */
	if (props > 39)
		return XZ_OPTIONS_ERROR;

	s->dict.size = 2 + (props & 1);
	s->dict.size <<= (props >> 1) + 11;

	if (DEC_IS_MULTI(s->dict.mode)) {
		if (s->dict.size > s->dict.size_max)
			return XZ_MEMLIMIT_ERROR;

		s->dict.end = s->dict.size;

		if (DEC_IS_DYNALLOC(s->dict.mode)) {
			if (s->dict.allocated < s->dict.size) {
				s->dict.allocated = s->dict.size;
				vfree(s->dict.buf);
				s->dict.buf = vmalloc(s->dict.size);
				if (s->dict.buf == NULL) {
					s->dict.allocated = 0;
					return XZ_MEM_ERROR;
				}
			}
		}
	}

	s->lzma.len = 0;

	s->lzma2.sequence = SEQ_CONTROL;
	s->lzma2.need_dict_reset = true;

	s->temp.size = 0;

	return XZ_OK;
}

XZ_EXTERN void xz_dec_lzma2_end(struct xz_dec_lzma2 *s)
{
	if (DEC_IS_MULTI(s->dict.mode))
		vfree(s->dict.buf);

	kfree(s);
}